routing loops on unify

Yes, we're in. Full path is working:

• Jump tunnel (localhost:9999 → clearsigint.com → fscwr) — had to restart it, it had dropped
• SSH to UDM (root@192.168.1.1) — connected fine
Gateway is RB-UDM-MAX-1010, UniFi OS 5.1.26, up 10 days 14 hours, load average a healthy 1.52 / 2.02 / 1.99 (much lower than yesterday's 3.7 — no tshark running). What do you need on it?

please some some diagnostics on this now please because the network is acting up...

Here's what I found. The good news is the gateway itself is healthy — but there's one clear culprit for the network acting up:

The Problem: tlprt0 Routing Loop

47 occurrences in dmesg (ongoing, not just boot-time):

wireguard: tlprt0: possible loop detected, dropping skb of size 65216

tlprt0 is the Teleport VPN tunnel. It's in a routing loop — packets destined for somewhere are being sent back into the Teleport interface, which detects the loop and drops them. This typically happens when:

• A Teleport client connects and its assigned IP range conflicts with an existing route
• Someone changed a Teleport network/subnet setting recently
• Notably: riccardo anselmo deleted 6 Teleport clients on 19 Aug — if new clients reconnected with conflicting routes, that could trigger this

Everything Else Looks Normal

Metric

Value

Status

Load

1.35 / 1.90 / 1.95 on 4 cores

✅ Normal

Temp

45°C

✅ Normal

Memory

1.8 GiB avail of 3.9 GiB

✅ Fine

Conntrack

2,963 / 65,536

✅ Fine

WAN eth2

24.89.146.106 UP, 0 err

✅ Up

WAN eth3

72.80.31.42 UP, 0 err

✅ Up

WAN eth6

199.219.255.66 UP, 4.5M drops

ℹ️ Multicast only (known)

WAN eth7

167.20.46.18 UP, 0 err

✅ Up

Ping 8.8.8.8

4ms, 0% loss

✅ Fine

Ping 1.1.1.1

8ms, 0% loss

✅ Fine

DNS

Resolving correctly

✅ Fine

WireGuard site-to-site

wgsts1000–1003 up

✅ Up

Also: ubios-udapi-server is flooding syslog with "No slot available for legacy unicast reflection" — this is mDNS/Avahi related, likely a device on the LAN spamming multicast queries. Not a direct cause of the network issue but worth watching.

What to Check

• Who is connected to Teleport? Go to UniFi → VPN → Teleport Clients and see if any recently-reconnected device has a subnet that overlaps with your LAN or site-to-site WAN pool.
• Disconnect/reconnect any active Teleport clients — the loop clears itself once the offending route is gone.
• The mDNS flood: if you can identify a specific device hammering multicast queries, that's worth isolating.
Want me to pull the Teleport client routing info directly from the UDM to pinpoint the conflicting subnet?